Harnessing Health Informatics: Dilemmas of Big Data and Adaptive Technologies

GLOBAL HEALTH WRITES

Citizen Journalist: Richard Gibney

Today, both patient confidentiality and data protection are highly valued. But so is the data itself. Data breaches in the health sector are a serious concern.  Many hospitals continue to rely on paper systems for their health records, with the move to digital seen as carrying risks.  Yet, these existing systems carry their own flaws.  Extreme examples include patient files found outside of the hospital in areas across Ireland, treated as general waste, left out before dumping, or discovered on roadsides.

Electronic data can also be easier to use in many ways.  Digitised forms of data are effectively shared and efficiently stored, saving on time, and space, and physical transport.  Systems of protection and encryption exist to ensure appropriate access and usage.  When it comes to the protection of data and information, where do we stand on raw anonymised patient data?

Incubating and Storing Data

Dr. Carolyn McGregor, Research Chair in Health Informatics at the University of Ontario Institute of Technology (UOIT) is heading a team of researchers on the Artemis project. The platform gathers and analyses biological information from premature babies that would have been unavailable just a few years ago; hundreds of time-sensitive data points, including respiration, heart rate, temperature, circulation.

McGregor first used this data to look into possible causes of infant fatalities. With their immune systems under-developed, premature infants (preemies) are more prone to infections. Upon examining vital indicators, like heartbeat regularity, McGregor and her team were able to inform predictions about infant survival rates. They found that preemies with acquired infections have more regular heartbeats than those without. This information was used to flag warning signs and inform future treatment for preemies struggling under similar circumstances.

Data analysts, statisticians and engineers, working alongside the medical profession, can now enhance incubator monitoring equipment using algorithms, based on these same key indicators provided by the data, so that the alarm bells are raised in a timely fashion. The result is life-saving intervention – such as blood transfusions to treat sepsis – for premature babies.

Streamlining and Time-Saving Data

The data can often represent thousands of pages of information when printed out. Analysed using computer technologies, the processes are far more efficient than conventional examination. Two decades ago, the storage capacity wouldn’t have been available on computer servers to retain the data for long periods. Monitors of many kinds usually recorded data for a given period, if at all, before being dumped from the system.

In real time, according to a report from IBM, if doctors or nurses manually checked the streaming data with all of these data points, and determined treatment for infection, the results would be hugely time-consuming and perhaps come too late. A huge amount of resources would be required for the monitoring of an incubator unit; watching for minute changes in temperature, heart rate, circulation or breathing problems. The monitoring extends to far more than just the baby’s heartbeat. For instance, if oxygen levels are too high in an incubator, a baby might suffer permanent eye damage.

In an email correspondence, Dr. McGregor conveyed that data analytics can be checked remotely through an extension of the platform, Artemis Cloud, between specialist centres thousands of miles apart, and communicated to local doctors and paediatricians if there are any signs of distress:

We started our work in China in 2012, in a 300 bed NICU at a hospital in Shanghai. We have run research students between my research lab at UOIT and the [Neonatal Intensive Care Unit] at Women and Infants Hospital in Providence, RI, USA and also between my research lab at UOIT and the hospital in Shanghai. We are using the strategic research computing network infrastructure in Ontario, called ORION, to connect another hospital, Southlake, to Artemis Cloud.”

Big Data in Health and Beyond

A single health centre might retain information on a vast number of vital and non-vital statistics on many hundreds or thousands of patients, from cell counts and pulse rates, to iron levels and skin conditions. In theory these could be mined to uncover subtle and nuanced patterns, and subsequent cures and treatments for a number of illnesses and health conditions.  However, until data is freed of identifiable patient markers, breaches in doctor-patient privilege will always be a risk. Adaptive technologies like the Artemis platform have massive benefits in health research with applications that extend far beyond the maternity wards.

In response to our queries about the future uses of data analytics beyond this field, Dr. McGregor wrote to us about the potential applications of Artemis, and where her work is currently:

There is research showing that you can use heart rate variability to determine whether to cease mechanical ventilation in [Intensive Care Unit] patients so as to reduce the incidence of the need for re-intubation. With such patterns known, we can use Artemis to watch for that pattern in real-time. There is [also] growing evidence to show that behaviours in physiological data like reduced heart rate variability can be associated with depression and post-traumatic stress disorder.

[With physiological data] I am also working on a new project called Athena which integrates data from a first person shooter computer game, ArmA 3, [using] a new form of haptic garment [a sensory-responsive suit], created by local startup ARAIG (As Real As It Gets)…We are using that to demonstrate new approaches for training tactical operators/SWAT teams.”

Richard Gibney is an editor for Yeah! International Student Magazine. You can follow both Yeah! and Richard on Twitter.

Email: gibneyrichard@gmail.com

Photo Credit: West Tennessee Healthcare Neonatal Intensive Care Unit: http://www.wth.org